Visual Sound Analysis

[Steve Herschbach]

18 hours ago, IBMe said:

Electronics has gone about as far as it can.

Signal analysis used to some extent.

Sound wave analysis, used in medical and industry, ignored.

Might be worth at least a little look.

 

I could be all wrong, but it’s used to tell the life of a bearing. I think it has strong possibilities.

I find the idea nobody has thought of this or looked at it as a possibility - ignored it - to be rather amusing. Of course it has been looked at and found wanting. Hook a signal analyzer up to any audio output of any metal detector and you will find out why. Test real world targets - lab demonstrations are fine and dandy but not how it works in the real world.

Hunting by ear can reveal differences people go by, but anyone that has dug enough targets knows bad sounding signals can be great finds, and great sounding targets can be junk. Part of this game is effiecient target recovery and over analyzing signals is generally counterproductive.

[Digalicious]

8 minutes ago, kac said:

If your machine has good modulated audio and it's recovery speed is relative to the size of a target then you should be able to hear the difference between a gold ring and piece of can slaw or pull tab. Aluminum will often have a sharper more abrupt sound and signal can carry further than a piece of gold with similar id  numbers.

 

Yet, the most experienced and knowledgeable jewelry hunters are digging massive amounts of aluminum trash. Why do you think that is Kac?

Just about any park or sports field will have a lot of gold rings. Problem is, who is going to dig all that aluminum trash to get those gold rings? I've heard of many who claim that they can distinguish between gold rings and aluminum trash. When asked to prove it with a video hunt in the wild, it's all crickets. 


 

[IBMe]

32 minutes ago, Digalicious said:

I agree, but that's a controlled test with very specific targets.

True, but some does translate to the field.

38 minutes ago, Digalicious said:

identical and overlapping

Not identical, mass density. Put a gold ring in an aluminum cap and another above. Put steel caps above and below. Test with and without the ring. You can easily tell there’s a small dense target in there. ID won’t be good.

How well that translates to field is another story.

[IBMe]

51 minutes ago, Steve Herschbach said:

I find the idea nobody has thought of this or looked at it as a possibility - ignored it - to be rather amusing.

A fairly newcomer amongst experts, how’s that going to turn out?

55 minutes ago, Steve Herschbach said:

Part of this game is effiecient target recovery and over analyzing signals is generally counterproductive.

Unless you’re 78 years old, then you might favor analyzation over digging.

1 hour ago, Steve Herschbach said:

Of course it has been looked at and found wanting. Hook a signal analyzer up to any audio output of any metal detector and you will find out why.

Sound analysis is used to tell when bearings need to be replaced. When they couldn’t find a vein in my arm, they brought in a sound machine with a display and found it.

None of that translates into metal detecting but may warrant further investigation.

Like I say, all I’m trying to do is make some noise holping for further experimentation. But I doubt the money is there. Most of the sales go to low-end machines. Medical and industrial use is very expensive.

 

[kac]

If I am in a park and I dig a flip tab that has a particular id and sound I am not going to dig every flip tab. People that dig all beat themselves up. I can't say that dig all is a successful way to hunt when we have machines that give us the ability to discriminate and id targets.

Dig all approach in a park is a good way to tear up the grounds and lose the chance of detecting there in the future. But then again people run around with full size spade shovels in parks and leave craters...

[Digalicious]

38 minutes ago, kac said:

If I am in a park and I dig a flip tab that has a particular id and sound I am not going to dig every flip tab. People that dig all beat themselves up. I can't say that dig all is a successful way to hunt when we have machines that give us the ability to discriminate and id targets.

Agreed:

When I'm using my Legend in an aluminum trash site and looking for gold, I don't dig 11, 46/47, and 28/29. For me, 11 has always been small foil, 46/47 has always been a penny, dime, or a full size aluminum screw cap, and 28/29 has always been a rectangular pull tab. 

Granted those lower numbers could be a gold ring, and the penny / dime signal could be a very large gold ring, but I play the odds...and the odds overwhelming tell me those numbers won't be a gold ring.

[kac]

The likely hood of a particular flip tab in an area having exactly the same response and id's as a ring is pretty remote.

I find it odd someone would keep digging the exact target type over and over again but in kindergarten we had craft time where we would doodle out something in crayon and use paste to make something of it and there was always one of those kids in the back eating the supplies.

[Digalicious]

15 hours ago, kac said:

The likely hood of a particular flip tab in an area having exactly the same response and id's as a ring is pretty remote.

 

Yes. The ID of 28/29 is always a "particular" type of tab for me, but depending on other variables, that particular tab can also ID a little higher or a little lower. Then of course, is all the other types of tabs.

To add insult to injury...foil / can slaw and all of its shapes and sizes 😡

[Chase Goldman]

I understand where @IBMe is coming from.  It's good to see the perspective and out-of-the-box ideas of newcomers to the hobby who are not biased by being steeped for years in the technologies and features that the detector manufacturers feed us.  Technology is not holding this idea back, because it has been tried before.  This is more about human biology (how humans process sensory input) and human factors engineering (how to present information in a manner that enables efficient and effective processing by a human) than it is about technology.  Furthermore, it probably makes more sense to directly "visualize" the processed target signal rather than the processed audio from that processed target signal.  And guess what, that has also been done with the various target trace implementations. 

The fact is, target audio is no accident, it is designed to provide the detectorist with a lot of target information that is embedded in the nuances and subtleties of the audio signal (volume, tonality, harmonics), and with repeated "training" and "muscle memory" via target recovery and audio conscious and subconscious signal association, enables the detectorist to learn those nuances and become even more effective than what can be visually displayed and interpreted on the fly.  That is not to say that visual target representations are not effective.  Of course they are, as evidenced by the more sophisticated target trace displays successfully used by detectorists on high end detectors.  But like I said, those are processed from the "source" target signal, rather than secondarily from the processed and generated audio waveforms.

Necessity is the mother of invention, as they say.  And the need to discover buried ordnance in the wars of the first half of the 20th century was the real the impetus for refining the induction balance and pulse induction metal detecting principles that are also used in the hobbyist detectors of today.  It is really defense and security applications, followed by gold prospecting that keep the technology progressing for the hobbyists.  If defense or security applications can be more effective with more sophisticated visual target representations, you can believe the detector manufacturers will invest and leverage that technology for hobbyist applications, if it is cost effective.

They key to "visual" target representation, whether you do that from the processed target signal directly or from the processed target audio, is determining how to effectively visualize the target attributes in a manner that enables the operator to enable the operator to efficiently, unambiguously, and reliably interpret those visual queues and to effectively integrate them with the audio target information they are also getting (without confusing the operator).  If the graphic visual target implementation can be picked up by the operator rather naturally with little need for "training" (i.e., swinging the detector for hundreds of hours before it "clicks") then that is a bonus. There have also been incremental advances in coil winding technology that have improved coil performance (depth and sensitivity) and ergonomics (weight).

Discrete tube-based circuits, gave way to semiconductor electronics, which fostered in more compact integrated circuits that enabled more sophisticated target processing in the form of rudimentary ferrous discrimination and ground effect compensation circuits as well as more sophisticated visual and audible target identification.  High speed digital signal processing and power electronics ushered in even more sophisticated target ID interfaces as well as enabling a greater ability to separate targets in high target density environments (high recovery speeds vs. depth), generate higher transmit powers with less weight and heat, enabled more sophisticated EMI cancellation, and also enabled advanced features like the myriad of simultaneous multiple frequency transmission protocols that exist today.

The next enabling technology for advancing the sophistication of both visual and audio target ID representations from where they are today probably resides in the detector designers' effective harnessing of the power of artificial intelligence/machine learning.  Imagine a detector that gets smarter/more effective at target identification (and has the ability to evolve how it presents that more effective data to the detectorist) the more you swing it.  That is probably the next frontier or evolution/revolution in advancing the state of the art in metal detection.

At least that's my take.

[Digalicious]

If an aluminum target is hit at 1 Khz and then 100 Khz, would the target signature change? If a gold target is hit at 1 Khz and then 100 Khz, would the target signature change?

What I'm getting at with the above, is if there is an amount of change in the signatures between the two, or one signature changes and the other doesn't, then can that amount of change (or lack thereof) be used to help differentiate between gold and silver?